【作者】 王晓哲；

【导师】 李从信；

【作者基本信息】 大庆石油学院 ， 计算机软件与理论， 2007， 硕士

【摘要】 经过近二十年的发展,游戏引擎已经逐渐成熟,尤其是3D游戏引擎。但由于诸多原因,我国对3D游戏引擎的研究相对滞后。国家已经将网络游戏引擎的开发纳入了863计划,这将大大推动我国3D游戏引擎开发的步伐。现在的3D游戏引擎越来越多地加强人工智能来提高游戏的可玩性。目前,智能角色已经成为3D游戏引擎中一个重要的部分,特别是对智能角色行为策略的研究,已成为一个研究热点。本文首先针对3D游戏引擎,介绍了其概念,分析并比较了几款著名的3D游戏引擎,结合全球著名的多功能的三维开发工具Virtools,对3D游戏引擎框架进行了深入地研究与分析,并利用Virtools软件开发了油田虚拟操作仿真演练系统。其次,介绍了3D游戏中的智能角色及其模型,总结了智能角色的行为特点,对智能角色的行为策略经行了深入研究。传统智能角色的开发方法是使用预先编辑的脚本来描述角色的行为,这种开发方法存在一定的缺陷。为了弥补这种缺陷,本文提出了适应性智能角色概念。Q学习算法总是选择最优策略,使用它设计的智能角色会一直处于最优状态。为了避免这种情况,引入了协调函数,对Q学习算法进行了改进。最后,利用改进的Q学习算法,在Virtools中,设计并实现了3D格斗游戏。经过多方面测试,3D格斗游戏中的适应性智能角色能够跟据与玩家的格斗战绩,自我调整智能水平,与玩家水平保持一致,克服了传统智能角色的缺陷,达到了预期的目的。更多还原

【Abstract】 Through the development of 20 years, the game engine,especially the 3D game engine, has already become mature. But because of many causes, the 3D game engine research was lagged in our country relatively. Our country has already brought the network game engine development into 863 plans, which quickens the step of 3D game engine development greatly in our country. Now the 3D game engine reinforces more artificial inllegence to make games more fun. The intelligent character plays a more and more important role in 3D game engine. Especially the research of intelligent character strategy has become a hot research problem.Firstly, the thesis has introduced the concept of 3D game engine, analyzed and compared some famous 3D game engines, made the thorough research and analysis of 3D game engine framework with Virtools which is the famous 3D develop tool in the world, and Virtools is applied in virtual operation simulation system. Secondly, it has introduced the intelligent character of 3D game engine and its model, summed up the behavior characteristic of intelligent character, and made the thorough research on the behavior strategy of intelligent character. The development means of traditional intelligent character is to describe the behavior of character with the pre-editing script. It has flaw. To rmedy this defect, , the thesis has brought forward the concept of the adaptability agent. Q_learning always choose an optimal strategy, the intelligence character who is designed by the strategy may always be in optimum state. To avoid this condition, it has led into the coordination function, and improved the Q_learning. Finally, It has designed and realizing the 3D fighting game in Virtools, using the improved Q_learning.By the many sides test, the adaptability agent of 3D fighting game can adjust the intelligence level by itself on the grounds of the military exploit, to keeping the level of player. It has overcome the defect of tradition agent, and has reached expected purpose consistently.更多还原